[bassman2914]

I plan to do a lot of aluminum work with my diy router if that has any effect on responses. I'm designing my router to be adjustable for the squareness of xy/yz/xz but I'm looking for values for what's good and what's not. The squareness of xy is likely the most important since most of what i'll be doing with be 2.5 d in aluminum plate. Also what is a good squareness for the spindle itself. By this I mean the linear rails can all be perfectly square so a given point on the cutting tool can move precisely but the spindle itself can be out of square with the actual direction of movement.(i'll make pics if that's hard to understand). My questions are what are good values for squareness for xy/xz/yz and spindle squareness with the rails in xz/yz planes. My machine will have a cutting area of 15x15x4 and will be used for work that is hopefully accurate within .001" in all dimension for anything i cut(mostly just for bragging rights, i will likely never "need" that precision). Thank You

[ger21]

You're parts will be as square as the machine. Any deviation from perpendicular in the X and Y axis will translate directly to your parts.

So a good value for squareness is "How square do you want your parts"?

[bassman2914]

What would you consider acceptable? with xy squareness I'll get it the best I can but adjusting the z's will be harder because of just how it'd have to be done and I don't want to fuss over trivial amount of accuracy. I understand it wont be perfect but i want it do be better mechanically than mach 3 could compensate for(considering it's not that great to begin with).

[danhamm]

The easiest way to level a z-axis is with a router or spindel installed, put a rod bent at right angles to the table, 1/4 in or 1/2 in in the chuck and rotate it using a feeler gauge or dial indicator, to the table..

[jsheerin]

I sweep a dial test indicator mounted in my spindle over a ~12" diameter circle and get within about +/- 0.001" front to back and side to side with my spindle. That lets me surface with a small cutter (1/4") and not be able to feel any ridge between overlapping cuts.

[vegipete]

In my opinion, this question points out a glaring deficiency in cnc controls. Why _should_ axes be orthogonal? If the control were better, it could be calibrated to 'misalignment' and compensate. A further example, why should it be necessary to dial a vise square onto a mill? Why not just plonk the vise anywhere onto the bed and 'calibrate' the control to the jaw face?

In the case of a mill, the only important alignment I can identify is that the spindle be as perfectly perpendicular as possible to the appropriate work piece surface. Sweeping a dial or test indicator around a suitably large circle, as jsheerin described above, is a very quick and effective to align the spindle.

[RICHARD ZASTROW]

vegipete, that technology has been available for decades. However, not offered much anymore. Probably not many takers. It's cheaper(?) than having an underpaid operator square it up.

Originated in the early CMM years.

Dick Z

[bassman2914]

Vegipete, both are probably possible through software but on a 3 axis router if the z isn't square you can never have a square part(a cube would have parallel but not perpendicular sides). With a 5 axis mill it's possible but less work is involved to square the machine than program the machine. If a jaw wasn't flat and you were trying to surface a 12" long part and the parts was out of square by.0005" per inch, then if your machine had an effective stepper resolution of .00025" (like mine) then then machine would make vertical steps every .5 inches of travel so it wouldn't be flat but would have a bunch of slopes that pop up .00025" every .5" to compensate for the angle. This is only if you are surfacing it relative to the part to keep metal removal to a minimum. Besides, the software can't compensate perfectly unless you have a VERY consistent machine with VERY high resolution with ZERO backlash. Not to be smart alec or anything.

[vegipete]

Don't get me wrong, I'm not saying a machine shouldn't be accurately built, but there are times when setup would be easier if the machine controls could zero to a part instead of zeroing the part to the machine. Indeed, on occasion I have clamped an odd or awkward piece into a mill, measured a couple a features to determine rotation and taken that info back to the cad software to adjust the model and generate g-code. That was easier than trying to position the part 'correctly.'

I disagree about Z-axis alignment. The spindle definitely needs to be square but the Z-axis can be crooked. The X and/or Y axes would move to compensate.

Certainly machine resolution would come into play for slight misalignment but the same effect will appear for any cuts that are almost parallel to an axis.

I'm not surprised that software axis alignment has been around but it does seem to me to be shame that it hasn't been made 'user friendly.' Axis parallelism calibration would only occur at machine commissioning and servicing, so the knuckledragger operator would be spared from that confusion. Hopefully said operator is not setting up the machine but if so, could software not be written so that measuring a vice position is easier than trusting him to adjust and properly tighten the vice? Would such features truly go unused in Mach or EMC for the home build crowd?

[bassman2914]

If the computer could adjust without with some reference points I would likely just use that because even though it's not as good and will be slightly inaccurate and leave "straight" lines less straight. In the average machining operation it likely wouldn't matter and it'd be much easier, faster, and increase productivity but since it's a pain and what i'm doing isn't hard to square up, i'll just stick to the old fashion way of adjusting it by hand. Back to my original question, how much does z-axis squareness matter in the average machining job with a hobby cnc where extreme accuracy a perfect fittings just isn't really required or easily possible. Mostly just opinion but don't just say the squarer the better.

[jsheerin]

My previous post is a good goal for how square the Z axis needs to be. I've had it less square than that and I get ridges between overlapping cuts that I can feel and see. If you did the above test and were out by +/-.005", with a 0.5" diameter cutter you'd have 0.0012" ridges between overlapping full width cuts. If you're working with wood, that might still be okay, but you can probably feel that on aluminum.